Pentatricopeptide-repeat family protein RF6 functionswith hexokinase 6 to rescue rice cytoplasmicmale sterilityWenchao Huanga,1, Changchun Yua,1, Jun Hua, Lili Wanga, Zhiwu Dana, Wei Zhoua, Chunlan Hea, Yafei Zenga,Guoxin Yaoa, Jianzhao Qia, Zhihong Zhanga, Renshan Zhua, Xuefeng Chena,b, and Yingguo Zhua,2

aNational Key Laboratory of Hybrid Rice, Key Laboratory for Research and Utilization of Heterosis in Indica Rice, Ministry of Agriculture, The Yangtze RiverValley Hybrid Rice Collaboration & Innovation Center, College of Life Sciences, Wuhan University, Wuhan 430072, China; and bThe Institute for AdvancedStudies, Wuhan University, Wuhan 430072, China

Edited by Maureen R. Hanson, Cornell University, Ithaca, NY, and accepted by the Editorial Board October 26, 2015 (received for review June 16, 2015)

Cytoplasmic male sterility (CMS) has been extensively used for hybridseed production in many major crops. Honglian CMS (HL-CMS) is oneof the three major types of CMS in rice and has contributed greatly tofood security worldwide. The HL-CMS trait is associated with anaberrant chimeric mitochondrial transcript, atp6-orfH79, which causespollen sterility and can be rescued by two nonallelic restorer-of-fertility (Rf) genes, Rf5 or Rf6. Here, we report the identificationof Rf6, which encodes a novel pentatricopeptide repeat (PPR) familyprotein with a characteristic duplication of PPR motifs 3–5. RF6 istargeted to mitochondria, where it physically associates with hexoki-nase 6 (OsHXK6) and promotes the processing of the aberrant CMS-associated transcript atp6-orfH79 at nucleotide 1238, which ensuresnormal pollen development and restores fertility. The duplicated mo-tif 3 of RF6 is essential for RF6-OsHXK6 interactions, processing of theaberrant transcript, and restoration of fertility. Furthermore, reduc-tions in the level of OsHXK6 result in atp6-orfH79 transcript accumu-lation and male sterility. Together these results reveal a novelmechanism for CMS restoration by which RF6 functions with OsHXK6to restore HL-CMS fertility. The present study also provides insightinto the function of hexokinase 6 in regulating mitochondrial RNAmetabolism andmay facilitate further exploitation of heterosis in rice.

cytoplasmic male sterility | pentatricopeptide repeat | restorer-of-fertility |RF6 | hexokinase 6

Plant cytoplasmic male sterility (CMS) is a trait characterizedby a lack of functional pollen grains in plants, although female

gametes are still viable (1). CMS has been widely used to producehybrid seeds to improve plant yield, improve resistance to diseasesor stresses, or enhance taste (2). Plant CMS results from genomicincompatibility between mitochondria and nuclei and is typicallyassociated with an aberrant chimeric gene in the mitochondria (3–6). A specific set of nuclear genes called restorer-of-fertility (Rf)genes, which primarily belong to the pentatricopeptide repeat(PPR) family, repress CMS through the promotion of the cleavage,degradation, or editing of CMS-associated aberrant transcripts orthe inhibition of the translation of these genes (7–9). PPR proteinsregulate RNA metabolism at multiple levels, from RNA editing,stability, processing, and splicing to translation (10, 11). PPR pro-teins can be classified into PLS and P families based on thestructure of PPRmotifs (11). PLS family members generally exhibitdirect RNA binding (12–14) or editing activity (15–17), whereasP-family proteins often lack sites for binding to RNA, particularlymitochondrial CMS-related RNA (18, 19). Most of the known PPRproteins implicated in restoration of fertility (RF) are P-familymembers. Thus, RF-related PPR proteins may require additionalpartners, such as RNA-binding or splicing factors to process CMS-related transcripts.Rice CMS is categorized into three major types: Wild Abortive

(WA), originating from a wild abortive rice line; Honglian (HL),originating from red-awned wild rice; and BT, originating fromChinsurahBoro II rice (20). These three types of CMS contribute

to ∼60% of the global hybrid rice production (21, 22). HL-CMS/Rfrice has been planted on over 6 million hectares in China andSoutheast Asia in the past 2 decades, significantly contributing tofood security worldwide (23). At the molecular level, WA-CMS isassociated with the mitochondrial gene WA352. Pollen abortion inthese rice plants occurs at the uninucleate stage and is suppressedby the nuclear genes Rf3 and Rf4 (24). BT-CMS involves the mi-tochondrial chimeric gene atp6-orf79 (25, 26). The pollen of theseplants aborts at the trinucleate stage and can be suppressed byRf1A and Rf1B (26). For HL-CMS, pollen aborts at the dinucleatestage and exhibits a spherical shape and negative staining withiodine-potassium iodide (I2-KI) solution (20). The HL-CMS trait isassociated with orfH79, a chimeric mitochondrial gene locateddownstream of atp6. The ORFH79 protein interacts with P61, asubunit of mitochondrial complex III, and impairs the activity ofthis complex. This inhibition leads to dysfunctional energy me-tabolism and elevated oxidative stresses in mitochondria, eventu-ally causing pollen sterility (27–30). The HL-CMS phenotype canbe rescued by two nonallelic Rf genes, Rf5 or Rf6 (31). F1 hybridscarrying a single restorer gene display ∼50% normal pollen grains,whereas those harboring both Rf genes generate 75% fertile pollenand a higher seed set rate under environmental stress conditions(31), thus indicating that both Rf genes are equally important for

Significance

Plant cytoplasmic male sterility (CMS) and sterility restoration bynuclear restorer-of-fertility (Rf) genes provide a crucial breedingtool to harness hybrid vigor in many crops. Here, we identify RF6as a novel pentatricopeptide repeat family protein that restoresthe fertility of Honglian CMS (HL-CMS), a major type of rice CMSused in breeding. We demonstrated that RF6 and hexokinase 6function together in mitochondria to promote the processing ofthe aberrant CMS-associated transcript atp6-orfH79, therebyrestoring the fertility of HL-CMS rice. This study links CMS res-toration to hexokinase, providing novel insights into the mech-anisms of CMS restoration, pollen development, mitochondriametabolism, and nuclear-cytoplasmic communication. In addi-tion, the present study offers opportunities to further exploitrice heterosis in hybrid rice production.

Author contributions: W.H. and Y. Zhu designed research; W.H., C.Y., J.H., L.W., Z.D., W.Z.,C.H., Y. Zeng, G.Y., and J.Q. performed research; W.H., Z.Z., X.C., and Y. Zhu analyzeddata; W.H., Z.Z., X.C., and Y. Zhu wrote the paper; and R.Z. provided the 9311 and YTA.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. M.R.H. is a guest editor invited by the EditorialBoard.

Freely available online through the PNAS open access option.1W.H. and C.Y. contributed equally to this work.2To whom correspondence should be addressed. Email: zhuyg@whu.edu.cn.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1511748112/-/DCSupplemental.

14984–14989 | PNAS | December 1, 2015 | vol. 112 | no. 48 www.pnas.org/cgi/doi/10.1073/pnas.1511748112

Dow

nloa

ded

by g

uest

on

Aug

ust 1

1, 2

020

maintaining fertility and the seed set rate. Recently, we havereported that Rf5 encodes a protein with 16 putative PPR motifsand that RF5 functions with GRP162, an RNA-binding protein, topromote the cleavage of the CMS-associated transcript orfH79 (18,32). In the present study, we identified Rf6 as a restorer gene forHL-CMS rice using a map-based cloning strategy. Mechanistically,we revealed that RF6 acts with hexokinase 6 (OsHXK6) in mito-chondria to reduce the accumulation of the CMS-associated tran-script atp6-orfH79 through a mechanism distinct from that of RF5.

ResultsIdentification of Rf6 as a Restorer Gene for HL-CMS Rice. We aimedto identify the other HL-CMS restorer gene, Rf6, using a map-based cloning strategy. In a previous study, we showed that Rf6cosegregates with the marker RM407 and associates with a regionof ∼200 kb flanked by the markers RM3710 and RM22242 onchromosome 8 (31) (Fig. 1A). To narrow down the location of Rf6,we selected an individual plant, (R-2), from the F2 populationobtained by crossing the restorer line (9311) with the HL-CMSabortive line (Yuetai A, or YTA) (31). R-2 contains YTA cyto-plasm and is homozygous at the Rf6 locus (Rf6Rf6), and 100% ofthe pollen grains generated from this plant exhibited black stainingwith 1% I2-KI solution. R-2 was subsequently backcrossed toYTA, generating a BC1F2 population containing 4,312 plants.Genetic linkage analysis identified two recombinant plants withthe RM3710 marker and three recombinant plants carrying theRM22242 marker from this population (Fig. 1A). To obtain ahigh-resolution map, we selected three polymorphic markers,RM407, ID200-1, and SNP32 (SI Appendix, Table S1), to screenfor individual recombinants. The markers RM407 and SNP32each identified a single recombinant plant, whereas the markerID200-1 did not (Fig. 1A). Thus, the Rf6 locus was delimited to a16.8-kb region between RM407 and SNP32 and cosegregated withthe marker ID200-1. In this region, there are two putative genes,AK066315 and AK068841, in the P0007D08 Nipponbare PACclone (Fig. 1A). To obtain the corresponding genomic sequence

for the restorer line, we constructed a 9311 BAC library (SI Ap-pendix, Fig. S1) and screened this library with the markers RM407and SNP32. A positive clone, 24C15, containing two putativeORFs, was identified (Fig. 1A) and subcloned for sequencing.Both ORFs (Orf1 and Orf2) were individually cloned into a vectorunder the control of the ubiquitin promoter of Zea mays andtransformed into the YTA sterile line for complementationanalyses. All 17 of the Orf1 transformants showed the same sterilephenotype as YTA. Conversely, all 19 of the transgenic T0 plantscontaining Orf2 displayed a normal fertile phenotype (Fig. 1B).Moreover, we observed that ∼50% of the pollen grains were fer-tile, as shown by I2-KI staining, and the seed-setting rate of theseplants was 80.5% on average (Fig. 1B and SI Appendix, Fig. S2).To confirm these results, three Orf2 T0 plants were self-crossed toproduce T1-generation plants. We observed a 1:1 ratio for plantswith 100% normal pollen compared with those with 50% abortivepollen in the T1 population (SI Appendix, Table S2), which indi-cates that Orf2 restores the fertility of HL-CMS in a gametophyticmanner. Thus, we demonstrated that Orf2 is the Rf6 gene re-sponsible for restoring HL-CMS fertility.

Duplicated 3–5 PPR Motifs Are Essential for Restoring HL-CMSFertility. Next, we used RT-PCR to examine the expression pat-terns of the Rf6 gene in R-2 plants and the recessive allele, rf6, inthe YTA abortive line. Both alleles exhibited similar expressionprofiles in a variety of organs (SI Appendix, Fig. S3), implying thatthe expression of these genes is not tissue-specific. Rf6 containsa mitochondrial transit signal at the N terminus (https://urgi.versailles.inra.fr/predotar/predotar.html). Indeed, we observed thata GFP-fused RF6 protein expressed from a plasmid driven by theCaMV 35S promoter in rice protoplasts was localized to mito-chondria (Fig. 2A). Sequence analysis revealed that Rf6 encodes anovel PPR protein of 894 amino acids with 20 PRR motifs (Fig.2B). Notably, the RF6 protein contained a duplication of motifs3–5, which are absent in rf6 or other known RF proteins (18, 33–38)(Fig. 2B and SI Appendix, Fig. S4). Outside this region, RF6 andrf6 are nearly identical in amino acid sequences (Fig. 2B), sug-gesting that this region is associated with the restoration of fer-tility. To evaluate the importance of each of the three PPR motifs,we synthesized a series of truncated Rf6 genes lacking these PPRmotifs (Fig. 2C, Rf6Ls and Rf6L-1 to Rf6L-9) and transformedthese genes into the YTA sterile line to examine their effects onfertility restoration (SI Appendix, Fig. S5A). We found that thedeletion of motif 3 alone or the deletion of motifs 3 and 4 or 3 and5 resulted in sterility in Rf6L-1, Rf6L-4, Rf6L-5, Rf6L-7, or Rf6L-8transgenic plants. However, the simultaneous disruption of motifs4 and 5 and retention of motif 3 (Rf6L-6 or Rf6L-9) did not impairfertility (Fig. 2C and SI Appendix, Fig. S5B and Table S2). Thus,the duplication of motif 3, but not 4 or 5, is essential and sufficientfor the function of Rf6 in fertility restoration. Furthermore, motiforder plays an important role in this process because the directfusion of motifs 5–3 (Rf6L-2, deletion of motif 4) resulted insterility (Fig. 2C and SI Appendix, Fig. S5B and Table S2), as motif4 alone is not required for fertility restoration. These three motifsdo not share a high degree of homology and are clearly classifiedinto distinct phylogenetic branches (SI Appendix, Fig. S6).Furthermore, to address whether this duplication could influ-

ence the structure of RF6, we used the crystal structure of maizechloroplast PPR10 (39) as a template to perform homology-basedmodeling of RF6, rf6, and RF6L-6 proteins. PPR10 belongs to theP-class of PPR family members and shares homology with RF6 inPPR motifs (SI Appendix, Fig. S7A). The duplicated motifs form aloosely extended structure connecting both N- and C-terminaldomains of the RF6 protein and become shorter in the RF6L-6and, particularly, the rf6 protein (SI Appendix, Fig. S7B). Wespeculated that this region may aid in establishing an optimal RF6conformation or serve as an interface for protein–protein inter-actions. To examine whether the duplicated region also appears in

Fig. 1. Rf6 restores the fertility of HL-CMS rice. (A) Schematic diagramshowing the mapping and location of Rf6. Clone number and restriction sitesare indicated. (B) Functional complementation analysis of Rf6 in rescuing HL-CMS sterility. Orf2 was transformed into the sterile line YTA (rf6rf6) byAgrobacterium tumefaciens-mediated methods. The anthers, 1% I2-KI–stainedpollen grains, and seed set rate for YTA and the transgenic plants carryingOrf2are shown. (Scale bars: 1 mm, 50 μm, and 500 mm in anthers, pollen grains, andseeds, respectively.)

Huang et al. PNAS | December 1, 2015 | vol. 112 | no. 48 | 14985

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 1

1, 2

020

other rice species, we amplified and analyzed Rf6 homologs from avariety of cultivars and their ancestors, the wild rice species withAA genomes (SI Appendix, Table S3). We observed that the du-plicated region also appeared in multiple indica cultivars and inseveral wild rice species (SI Appendix, Fig. S8). Thus, the dupli-cated region in RF6 likely originated from a duplication event thatoccurred in wild rice species.

RF6 Promotes Processing of the Chimeric Transcript atp6-orfH79 atNucleotide 1238. Based on previous studies, the sterility of HL-CMS rice is caused by the presence of the ORFH79 proteinproduct from the chimeric transcript atp6-orfH79 in mitochon-dria. ORFH79 impairs the function of the mitochondrial elec-tron transport chain complex III through an association with P61

(27–30). The restorer protein RF5 and its cofactor, GRP162,promote the cleavage of this aberrant transcript into twofragments, atp6 and orfH79, leading to the disruption of theORFH79 protein and fertility restoration (18). Thus, we exam-ined whether Rf6 restores the fertility of abortive lines throughthe same mechanism. The atp6-orfH79 transcript was largelyreplaced by two separate fragments, atp6 and orfH79, in trans-genic plants carrying Rf6, whereas this transcript remained intactin YTA, which carries rf6 (Fig. 3 A and B and SI Appendix, Fig.S9A). Consequently, the ORFH79 protein was not detected inthe presence of Rf6 (Fig. 3C). Furthermore, we observed theaccumulation of the atp6-orfH79 transcript and OFRH79 proteinin transgenic plants without motif 3 duplications (SI Appendix,Table S2 and Fig. S9B). These observations demonstrated thatRf6 and motif 3 are crucial for preventing the accumulation ofthe abnormal transcript. Next, we further defined the transcriptchanges associated with fertility restoration. We performed aprimer extension assay using Rf6 T0 transgenic plants and ob-served that a specific band appeared 12 nucleotides (at the Gresidue) downstream from the start codon of orfH79 but was notdetected in YTA (Fig. 3D). This observation likely reflects thecleavage of the chimeric transcript by RNA-processing factors.However, we cannot fully exclude the possibility that the restorercomplex may protect the transcript against exonuclease pro-cessing or block the translation of the transcript. This processingsite is located at nucleotide 1238 on a single-stranded loop in theatp6-orfH79 transcript (Fig. 3E) and obviously differs from theone generated by RF5, which lies in the junction region between

Fig. 2. The duplicated PPR motifs 3–5 in RF6 are crucial for restoring fer-tility. (A) Localization of Rf6 in mitochondria. Protoplasts transformed with35S:GFP:nos (CK) and 35S:Rf6-GFP: nos (Rf6) constructs were analyzed usingfluorescence microscopy. The dye Mito-Tracker Red was used as a mito-chondrial marker. (Scale bars: 10 μm.) (B) Comparison of the amino acidsequences for the RF6 and rf6 proteins. Sequence alignment was performedusing DNAMAN6.0 software. The PPR motifs are indicated with solid lines.SP: signal peptide. (C) Schematic diagrams showing wild-type Rf6 andengineered Rf6L constructs (Rf6L-1–Rf6L-9). The crosses involved the de-letion of the corresponding PPR motifs. F: fertility; S: sterility.

Fig. 3. RF6 promotes the processing of the atp6-orfH79 transcript at nu-cleotide 1238. (A) The schematic diagram of the atp6-orfH79 chimeric geneand transcript. (B) Northern blot analysis showing the level of the atp6-orfH79 transcript in YTA and the transgenic plant with Rf6. The orfH79-coding sequence was used as a probe. (Lower) A gel stained with ethidiumbromide to confirm that equal amounts of RNA were loaded in each lane.(C) A Western blot showing the protein level of ORF79 in transgenic plantswith Rf6. YTA and R-7 were used as negative and positive controls, re-spectively. HSP90 protein was used as a loading control. (D) A primer ex-tension assay mapping the processing site in the atp6-orfH79 transcript in T0transgenic plants. The processing site G (red) is indicated with an arrow.(E) The secondary structure of the atp6-orfH79 transcript was predicted by theMfold Web Server (unafold.rna.albany.edu). The processing site at nucleotide1238 of the atp6-orfH79 transcript is indicated with an arrow. The RNA andprotein for these experiments were prepared from plant leaves.

14986 | www.pnas.org/cgi/doi/10.1073/pnas.1511748112 Huang et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 1

1, 2

020

atp6 and orfH79 at nucleotide 1169 of the same transcript (18).Thus, RF6 likely differs from RF5 mechanistically in processingthe atp6-orfH79 transcript.

RF6 Does Not Bind the atp6-orfH79 Transcript or GRP162 Protein. Tounderstand how Rf6 may process the aberrant atp6-orfH79 tran-script, we examined whether Rf6 can directly bind to the transcript.We purified the RF6 protein using an maltose-binding protein tag(SI Appendix, Fig. S10A) and incubated this fusion protein with abiotin-labeled RNA probe containing the linker region betweenthe atp6- and orfH79- coding sequences (Fig. 3A). We did notobserve any shift in electrophoretic mobility shift assays (EMSAs),even when RF6 was added at a concentration of 10 nM (SI Ap-pendix, Fig. S10B), which suggests that RF6 does not directly bindto the atp6-orfH79 transcript. RF5 functions through an associationwith GRP162, a glycine-rich protein that binds to atp6-orfH79mRNA, to promote the processing of the aberrant transcript (18).Therefore, we examined whether RF6 interacts with GRP162 by ayeast two-hybrid assay. However, we did not observe any signs ofinteraction between these proteins (SI Appendix, Fig. S11), sug-gesting that RF6 does not bind to this RNA-processing factor.Thus, RF6 likely requires other types of cofactors for binding andprocessing the atp6-orfH79 transcript.

RF6 Physically Interacts with OsHXK6 Both in Vitro and in Vivo. Toidentify potential targets that interact with RF6, we generated ayeast two-hybrid library for F1 hybrids and screened this libraryusing RF6 protein as bait. Among the proteins identified (SIAppendix, Table S4), OsHXK6, a putative rice hexokinase local-ized to the mitochondria and nucleus (40), was detected. Weobserved that OsHXK6 displayed an expression pattern similar tothat of RF6 and was primarily distributed within the insolublefraction of mitochondria (Fig. 4A and SI Appendix, Fig. S13A).Importantly, OsHXK6 exhibited a strong specific interaction withRF6 but not rf6 in the yeast two-hybrid assay (Fig. 4B). This resultwas confirmed by a pull-down assay with purified GST-OsHXK6

and MBP-RF6 or MBP-rf6 proteins (Fig. 4C and SI Appendix, Fig.S10A). Furthermore, we performed a coimmunoprecipitation as-say using purified mitochondria from plants with Myc-tagged RF6and observed a strong interaction between OsHXK6 and RF6 inmitochondria (Fig. 4D). Thus, we demonstrated that RF6 directlyinteracts with OsHXK6 both in vitro and in vivo and that theduplicated motifs are essential for this interaction.

C terminus of RF6 Mediates the RF6–OsHXK6 Interaction. Next, wesought to identify the RF6 domains that directly mediate the as-sociation with OsHXK6. We examined the association of OsHXK6with truncated RF6 proteins by a yeast two-hybrid assay. Surpris-ingly, the region that directly interacted with OsHXK6 was locatedat the C terminus of RF6, from residue 386 to 894, whereas theduplicated motifs at the N terminus did not interact with OsHXK6(SI Appendix, Fig. S12A). Thus, these results suggest that du-plicated motifs are needed to establish an optimal conformationappropriate for OsHXK6 association rather than serving as adirect docking site.

Association with RF6 Does Not Depend on the Catalytic Motif ofOsHXK6. Two mutations in the Gly (G112) residue to Asp (D) andin a Ser (S185) residue to Ala (A) at the N terminus of OsHXK6result in the loss of sugar-kinase catalytic activity in rice (40). Thus,we examined whether the catalytic domain of OsHXK6 is requiredfor association with RF6. To this end, we performed a yeast two-hybrid assay using full-length RF6 as bait and truncated OsHXK6as prey. We observed strong interactions between RF6 and trun-cated OsHXK6, encompassing residues 386–506. However, we didnot observe a direct association between RF6 and the OsHXK6N-terminal domain, which included the sugar-kinase catalytic motif(SI Appendix, Fig. S12B). This result suggests that the associationwith RF6 does not involve the catalytic domain of OsHXK6.

OsHXK6 Promotes the Processing of the atp6-orfH79 Transcript andRestoration of Fertility. The above results suggest that RF6 andOsHXK6 may function together to process the atp6-orfH79 tran-script and restore fertility. To examine the role of OsHXK6 in thisprocess, we used RNA interference to down-regulate the expres-sion of the OsHXK6 gene in R-7 F3 plants, which are homozygousat the Rf6 locus and exhibit 100% normal pollen grains. Weobtained three transgenic lines with reduced levels of OsHXK6 (SIAppendix, Fig. S13B). As expected, the atp6-orfH79 transcript ac-cumulated in OsHXK6i plants (SI Appendix, Table S2), andORFH79 protein also accumulated (Fig. 4E). As a result, we ob-served that ∼50% of pollen grains were abortive in these plants(Fig. 4F and SI Appendix, Fig. S13 C and D), a proportion equiv-alent to that in the F1 hybrids. Therefore, we concluded that RF6functions with OsHXK6 in mitochondria to reduce the accumu-lation of the atp6-orfH79 transcript, thus leading to the restorationof HL-CMS fertility. Furthermore, we observed that the expressionof the Rf6 gene inOsHXK6i transgenic lines remains unaffected (SIAppendix, Fig. S14), suggesting that the roles of OsHXK6 in re-storing fertility are not through impacting the expression of Rf6.

DiscussionIn the present study, we report the identification and character-ization of a novel PPR family restorer gene, Rf6, and demonstratethat RF6 protein works with OsHXK6 in mitochondria to stimulatethe processing of the HL-CMS–associated transcript, atp6-orfH79,leading to normal pollen development and fertility restoration. Theresults suggest that RF6 and RF5 function through distinct mech-anisms to rescue the sterility of HL-CMS. Unlike RF5, RF6 doesnot interact with the RNA-processing factor, GRP162 (SI Appendix,Fig. S11). Additionally, RF6 likely promotes the cleavage of theatp6-orfH79 transcript at a novel position, as shown in the model(Fig. 5). To our knowledge, this study is the first to link restorationof CMS fertility to the hexokinase OsHXK6, providing novel

Fig. 4. OsHXK6 functions with RF6 to promote the restoration of HL-CMSfertility. (A) A Western blot showing the coexistence of the RF6 and OsHXK6proteins in insoluble mitochondria fractions. (B) Yeast two-hybrid analysisresults showing the interaction between OsHXK6 and RF6 proteins. The 53-Lam and 53-T interactions were used as negative and positive controls, re-spectively. DDO: SD-Leu/-Trp medium; QDO: SD/-Leu/-Trp/-His/-Ade medium;X: X-αIL. (C) A pull-down assay revealing the direct physical interaction be-tween OsHXK6 and RF6 but not rf6 in vitro. A “+” and a “−” indicate thepresence and absence of corresponding components in the reactions, re-spectively. (D) Coimmunoprecipitation of the RF6 and OsHXK6 proteins frommitochondria. (E) Western blot analysis of the ORFH79 protein level in in-dicated plants. YTA and R-7 were used as negative and positive controls,respectively. HSP90 protein was used as a loading control. (F) Microscopicanalysis of pollen grains stained with 1% I2-KI solution in the indicatedplants. (Scale bars: 50 μm.)

Huang et al. PNAS | December 1, 2015 | vol. 112 | no. 48 | 14987

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 1

1, 2

020

insights into the mechanisms of CMS restoration, pollen devel-opment, and hexokinase 6 function in regulating mitochondrialRNA metabolism.The PPR protein family is one of the most perplexing protein

families in plants such as maize, rice, sorghum, rapeseed, petu-nia, radish, and sunflower (9). PPR proteins are tandem arrays of2–27 repeats of PPR motifs, and each motif contains 35 aminoacids (7). Structurally, PPR motifs consist of two distinct anti-parallel alpha helices, helices A and B, closely resembling thetetratricopeptide repeat motif, which is involved in mediatingprotein–protein interactions (7, 41). We observed that RF6contains a unique duplication of PPR motifs, which are notpresent in rf6 or other known RF proteins. This duplication doesnot interact with OsHXK6 but is essential for enabling RF6 tointeract with OsHXK6 to process the atp6-orfH79 transcript.Thus, this duplication likely establishes an optimal conformationappropriate for OsHXK6 association or is associated with anunknown mediator protein in the restoration complex.The HL-CMS restorer gene, Rf5, which encodes a P-family

member of PPR family proteins, is identical to the BT-CMS re-storer gene, Rf1A (18). However, Rf1A directly binds to and pro-cesses the CMS-associated transcript atp6-orf79 in the BT-CMS

line (42), whereas Rf5 apparently requires additional factors, in-cluding GRP162, to process the aberrant transcript atp6-orfH79 inHL-CMS lines (18), suggesting that RF5 protein binds both proteinand RNA molecules, depending on the specific situation in a givenCMS system. Importantly, we observed that Rf6 also restored thefertility of BT-CMS rice. Therefore, it will be necessary to testwhether Rf6 functions through the same mechanism in the twoCMS systems and whether OsHXK6 is needed to rescue BT-CMSsterility. Regardless, a broad role for Rf6 in rescuing both HL- andBT-CMS sterility should greatly facilitate the further exploitationof heterosis in rice breeding.In addition to functioning as glucose sensors, hexokinase

proteins phosphorylate hexose to form hexose-6-P, thereby regu-lating many physiological processes, such as germination, flower-ing, senescence, and pathogen defense responses to biotic andabiotic stresses in plants (43–45). OsHXK6 belongs to a hexoki-nase family comprising 10 members in rice (46). Among thesemembers, OsHXK6 was closely related to AtHXK1, which func-tions in a complex involved in regulating gene expression or RNAmetabolism in the nucleus (47). OsHXK6 can be exported into thenucleus, and the nuclear localization of this protein may be criticalfor its sugar-mediated signaling functions in rice (40). In contrast,little is known regarding the function of OsHXK6 in mitochon-dria. Here, we identified OsHXK6 as a RF6 partner that partici-pates in the processing of a mitochondrial RNA transcript. Thesugar-kinase catalytic domain is not involved in interactions withRF6, suggesting that the function of OsHXK6 in fertility resto-ration is likely independent of its role in sugar sensing. However,the precise role of OsHXK6 in this process is unclear. Althoughboth RF6 and OsHXK6 are essential for processing the chimerictranscript, neither protein can directly bind to RNA (SI Appendix,Figs. S10B and S15), implying that additional proteins capable ofbinding or splicing RNA should be involved (Fig. 5). Identifyingand characterizing the functions of these factors will be an im-portant task necessary for fully elucidating the mechanism of CMSand fertility restoration. In addition, this study provides informa-tion for further studies on the function of hexokinases in mito-chondria metabolism and nuclear-cytoplasm communication.

Materials and MethodsInformation on the plant materials used in this study is included in SI Ap-pendix, Materials and Methods. Total RNA was extracted using TRIzol re-agent (Invitrogen). The yeast two-hybrid (Clontech) and EMSA (ThermoScientific) assays were performed according to the manufacturer’s instruc-tions. Detailed experimental procedures and reagents used in this study aredescribed in SI Appendix, Materials and Methods.

ACKNOWLEDGMENTS. We thank Dr. Jianming Wan, Dr. Haiyang Wang,Dr. William Terzaghi, Dr. Marenda Wilson, and Dr. Mengxiang Sun for criticalcomments on the manuscript and Dr. Meizhong Luo for providing BAC vectors.This work was supported by grants from the National Natural ScienceFoundation of China (Grant 31371236) and the National High TechnologyResearch and Development Program of China (Grant 2011AA10A101).

1. Chen L, Liu YG (2014) Male sterility and fertility restoration in crops. Annu Rev PlantBiol 65:579–606.

2. Havey M (2004) The use of cytoplasmic male sterility for hybrid seed production.Molecular Biology and Biotechnology of Plant Organelles, eds Daniell H, Chase CD(Springer, Berlin), pp 623–634.

3. Levings CS, III (1993) Thoughts on cytoplasmic male sterility in cms-T maize. Plant Cell5(10):1285–1290.

4. Grelon M, Budar F, Bonhomme S, Pelletier G (1994) Ogura cytoplasmic male-sterility(CMS)-associated orf138 is translated into a mitochondrial membrane polypeptide inmale-sterile Brassica cybrids. Mol Gen Genet 243(5):540–547.

5. Schnable PS, Wise RP (1998) The molecular basis of cytoplasmic male sterility andfertility restoration. Trends Plant Sci 3:175–180.

6. Balk J, Leaver CJ (2001) The PET1-CMSmitochondrial mutation in sunflower is associatedwithpremature programmed cell death and cytochrome c release. Plant Cell 13(8):1803–1818.

7. Small ID, Peeters N (2000) The PPR motif: A TPR-related motif prevalent in plant or-ganellar proteins. Trends Biochem Sci 25(2):46–47.

8. Dahan J, Mireau H (2013) The Rf and Rf-like PPR in higher plants, a fast-evolvingsubclass of PPR genes. RNA Biol 10(9):1469–1476.

9. Barkan A, Small I (2014) Pentatricopeptide repeat proteins in plants. Annu Rev PlantBiol 65:415–442.

10. Lurin C, et al. (2004) Genome-wide analysis of Arabidopsis pentatricopeptide re-peat proteins reveals their essential role in organelle biogenesis. Plant Cell 16(8):2089–2103.

11. Schmitz-Linneweber C, Small I (2008) Pentatricopeptide repeat proteins: A socket setfor organelle gene expression. Trends Plant Sci 13(12):663–670.

12. Schmitz-Linneweber C, Williams-Carrier R, Barkan A (2005) RNA immunoprecipitationand microarray analysis show a chloroplast Pentatricopeptide repeat protein to beassociated with the 5′ region of mRNAs whose translation it activates. Plant Cell17(10):2791–2804.

13. Beick S, Schmitz-Linneweber C, Williams-Carrier R, Jensen B, Barkan A (2008) Thepentatricopeptide repeat protein PPR5 stabilizes a specific tRNA precursor in maizechloroplasts. Mol Cell Biol 28(17):5337–5347.

Fig. 5. A working model showing the mechanism by which RF6 restores thefertility of HL-CMS. RF6 and OsHXK6 proteins are targeted to mitochondria,forming a complex with other proteins, including RNA-binding and processingfactors, likely to cleave the aberrant atp6-orfH79 transcript, thus disruptingorfH79 mRNA and preventing its translation. Consequently, this processing en-sures normal pollen development and leads to the restoration of CMS fertility.

14988 | www.pnas.org/cgi/doi/10.1073/pnas.1511748112 Huang et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 1

1, 2

020

14. Pfalz J, Bayraktar OA, Prikryl J, Barkan A (2009) Site-specific binding of a PPR proteindefines and stabilizes 5′ and 3′ mRNA termini in chloroplasts. EMBO J 28(14):2042–2052.

15. Kotera E, Tasaka M, Shikanai T (2005) A pentatricopeptide repeat protein is essentialfor RNA editing in chloroplasts. Nature 433(7023):326–330.

16. Okuda K, et al. (2010) The pentatricopeptide repeat protein OTP82 is required forRNA editing of plastid ndhB and ndhG transcripts. Plant J 61(2):339–349.

17. Robbins JC, Heller WP, Hanson MR (2009) A comparative genomics approach iden-tifies a PPR-DYW protein that is essential for C-to-U editing of the Arabidopsischloroplast accD transcript. RNA 15(6):1142–1153.

18. Hu J, et al. (2012) The rice pentatricopeptide repeat protein RF5 restores fertility inHong-Lian cytoplasmic male-sterile lines via a complex with the glycine-rich proteinGRP162. Plant Cell 24(1):109–122.

19. Gillman JD, Bentolila S, Hanson MR (2007) The petunia restorer of fertility proteinis part of a large mitochondrial complex that interacts with transcripts of the CMS-associated locus. Plant J 49(2):217–227.

20. Li SQ, Yang DC, Zhu YG (2007) Characterization and use of male sterility in hybrid ricebreeding. J Integr Plant Biol 49:791–804.

21. Virmani SS (2003)Advances in hybrid rice research and development in the tropics. InProceedings of the 4th International Symposium on Hybrid Rice, eds Virmani SS,Mao CX, Hardy B (International Rice Research Institute, Manila, Philippines), pp 7–20.

22. Cheng SH, Zhuang JY, Fan YY, Du JH, Cao LY (2007) Progress in research and devel-opment on hybrid rice: A super-domesticate in China. Ann Bot (Lond) 100(5):959–966.

23. Huang W, et al. (2012) Research and development of HL type hybrid rice. Sci ChinaLife Sci 42(9):689–698.

24. Luo D, et al. (2013) A detrimental mitochondrial-nuclear interaction causes cytoplasmicmale sterility in rice. Nat Genet 45(5):573–577.

25. Akagi H, Sakamoto M, Shinjyo C, Shimada H, Fujimura T (1994) A unique sequencelocated downstream from the rice mitochondrial atp6 may cause male sterility. CurrGenet 25(1):52–58.

26. Wang Z, et al. (2006) Cytoplasmic male sterility of rice with boro II cytoplasm is causedby a cytotoxic peptide and is restored by two related PPR motif genes via distinctmodes of mRNA silencing. Plant Cell 18(3):676–687.

27. Yi P, Wang L, Sun Q, Zhu YG (2002) Discovery of mitochondrial chimeric gene asso-ciated with male sterility of HL-rice. Chin Sci Bull 47(9):744–747.

28. Peng X, et al. (2010) The mitochondrial gene orfH79 plays a critical role in impairingboth male gametophyte development and root growth in CMS-Honglian rice. BMCPlant Biol 10:125.

29. Liu G, et al. (2012) Alterations of mitochondrial protein assembly and jasmonic acidbiosynthesis pathway in Honglian (HL)-type cytoplasmic male sterility rice. J Biol Chem287(47):40051–40060.

30. Wang K, et al. (2013) ORFH79 impairs mitochondrial function via interaction with asubunit of electron transport chain complex III in Honglian cytoplasmic male sterilerice. New Phytol 198(2):408–418.

31. Huang W, et al. (2012) Two non-allelic nuclear genes restore fertility in a gametophyticpattern and enhance abiotic stress tolerance in the hybrid rice plant. Theor Appl Genet124(5):799–807.

32. Hu J, et al. (2013) The mechanism of ORFH79 suppression with the artificial restorerfertility gene Mt-GRP162. New Phytol 199(1):52–58.

33. Bentolila S, Alfonso AA, Hanson MR (2002) A pentatricopeptide repeat-containinggene restores fertility to cytoplasmic male-sterile plants. Proc Natl Acad Sci USA99(16):10887–10892.

34. Brown GG, et al. (2003) The radish Rfo restorer gene of Ogura cytoplasmic malesterility encodes a protein with multiple pentatricopeptide repeats. Plant J 35(2):262–272.

35. Akagi H, et al. (2004) Positional cloning of the rice Rf-1 gene, a restorer of BT-typecytoplasmic male sterility that encodes a mitochondria-targeting PPR protein. TheorAppl Genet 108(8):1449–1457.

36. Komori T, et al. (2004) Map-based cloning of a fertility restorer gene, Rf-1, in rice(Oryza sativa L.). Plant J 37(3):315–325.

37. Klein RR, et al. (2005) Fertility restorer locus Rf1 [corrected] of sorghum (Sorghumbicolor L.) encodes a pentatricopeptide repeat protein not present in the colinearregion of rice chromosome 12. Theor Appl Genet 111(6):994–1012.

38. Tang H, et al. (2014) The rice restorer Rf4 for wild-abortive cytoplasmic male sterilityencodes a mitochondrial-localized PPR protein that functions in reduction of WA352transcripts. Mol Plant 7(9):1497–1500.

39. Yin P, et al. (2013) Structural basis for the modular recognition of single-strandedRNA by PPR proteins. Nature 504(7478):168–171.

40. Cho J-I, et al. (2009) Role of the rice hexokinases OsHXK5 and OsHXK6 as glucosesensors. Plant Physiol 149(2):745–759.

41. Das AK, Cohen PW, Barford D (1998) The structure of the tetratricopeptide repeats ofprotein phosphatase 5: Implications for TPR-mediated protein-protein interactions.EMBO J 17(5):1192–1199.

42. Kazama T, Nakamura T, Watanabe M, Sugita M, Toriyama K (2008) Suppressionmechanism of mitochondrial ORF79 accumulation by Rf1 protein in BT-type cyto-plasmic male sterile rice. Plant J 55(4):619–628.

43. Jang JC, León P, Zhou L, Sheen J (1997) Hexokinase as a sugar sensor in higher plants.Plant Cell 9(1):5–19.

44. Cho J-I, Ryoo N, Hahn T-R, Jeon J-S (2009) Evidence for a role of hexokinases asconserved glucose sensors in both monocot and dicot plant species. Plant SignalBehav 4(9):908–910.

45. Bolouri-Moghaddam MR, Le Roy K, Xiang L, Rolland F, Van den Ende W (2010) Sugarsignalling and antioxidant network connections in plant cells. FEBS J 277(9):2022–2037.

46. Cho JI, et al. (2006a) Structure, expression, and functional analysis of the hexokinasegene family in rice (Oryza sativa L.). Planta 224(3):598–611.

47. Cho Y-H, Yoo S-D, Sheen J (2006) Regulatory functions of nuclear hexokinase1complex in glucose signaling. Cell 127(3):579–589.

Huang et al. PNAS | December 1, 2015 | vol. 112 | no. 48 | 14989

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 1

1, 2

020